This blog is a about a project we did for a class at the University of Salzburg, under the supervision of Dr. Euro Beinat. The main component is the analysis of tweets in certain parts of europe (more on that subject later on), concerning the topic of money.
We basically were given all the geolocated data in the region of europe for the timespan of february 2012. For our project only tweets with geographical coordinates
are interresting. Approximately 5 to 10 % of all the twitter data we got had
this coordinates we needed, nevertheless it was enough data to start our
project.
In this Blog we'll present the most important steps of the workflow and of course some results. This includes for example a data-review. Where are tweet intensive spots in europe? Where do the tweets come from? We picked the special topic of "Money-Talk" - or in other words - how much do the people talk about money and financial matters on twitter? And if they tweet about it, is it more in the positive "I-won-the-lottery-way", or do they use twitter to complain about their financial
sorrows?
Recent research suggests that very early indicators about the status of the stock market can
be extracted from online social media including blogs and Twitter feeds (Bollen et al 2011).
Behavioral finance has provided proof that people do not make financial decisions in isolation,
they make decisions while interacting with others in social context and therefore the social
mood governs the tone and the character of financial and economic activity (Nofsinger 2003). Some researchers have used mood tracking tools to analyse large volumes of twitter feeds in order
to deceive the public mood and then use the the information to measure the
influence of public moods on financial stock markets (Bollen et al 2011, Mittal and Goel 2011).
Due to time issues we won't really be able to check if there is any "real" correlation between the financial mood in our selected twitter-dataset and the actual happenings at the stock market. Nevertheless we found the idea very interesting - thats why we decided to do at least this "mood-analysis" for our dataset.
On the following Blog - pages you'll find our ways of exploring the data and of course some of the analysis - including some results - which we did on the twitter dataset.
